The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download the Alibaba Cloud CLI and installation scripts from official vendor domains (aliyuncli.alicdn.com) and the vendor's GitHub organization (github.com/aliyun). The instructions emphasize security by recommending users inspect scripts before execution and providing methods to install the CLI into user-writable directories without requiring root privileges.
[COMMAND_EXECUTION]: The skill interacts with cloud services through the aliyun CLI. It implements a mandatory 'Parameter Confirmation Gate' which requires the agent to output a checklist of all parameters and wait for user confirmation before executing commands. This ensures human-in-the-loop oversight for all cloud API interactions. Additionally, it provides detailed error handling and retry logic to prevent silent failures.
[DATA_EXFILTRATION]: The skill documentation includes strict prohibitions against accessing or printing sensitive credentials (Access Keys). It provides clear instructions on checking credential status using non-sensitive commands like aliyun configure list and emphasizes the use of environment variables or secure local configuration files.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it processes monitoring reports and telemetry data from cloud APIs.
Ingestion points: Data enters the context through aliyun ebs describe-metric-data and aliyun ebs get-report outputs.
Boundary markers: The skill relies on structured JSON processing using jq and --cli-query to isolate data, reducing the likelihood of the agent misinterpreting data as instructions.
Capability inventory: The skill can execute read-only cloud queries via the CLI but lacks the capability to perform destructive operations or unauthorized network transfers.
Sanitization: External data is handled as structured objects rather than raw strings within prompts.

alibabacloud-ebs-usage-summary